TESTING IN A CONTINUOUS DELIVERY WORLD Improve speed without rushing software out the door.

Transcription

1 TESTING IN A CONTINUOUS DELIVERY WORLD Improve speed without rushing software out the door. Testing needs to shift left : Testing is starting to be done by developers more frequently. QA professionals are still doing manual work, but they re trying to automate the process as well. Challenge for Testers: not just to be a good tester but also be able to engineer the process and take advantage of advanced automation practices. KEY AREAS IN CONTINUOUS TESTING Risk Assessment. Defect Casual Analysis. Code Quality Control. Traceability. Test Optimization. Service Virtualization. KEY BENEFITS Focus on the areas that matter Determine current gaps in maturity Control risks, quality and costs

2

3 BASIC TESTING AWARE OF RISKS MEASURED QUALITY MEASURED COSTS EFFICIENT TESTING CONTROLLED RISKS CONTROLLED QUALITY CONTROLLED COSTS CONTINUOUS TESTING REDUCED RISKS OPTIMIZED QUALITY OPTIMIZED COSTS SOURCE CODE Source code versioning Code quality control CONTINUOUS INTEGRATION / CONTINUOUS DELIVERY ENVIRONMENT / INFRASTRUCTURE Separted /dev environments Few devices/browsers are covered Test data management Full devices/browsers coverage Virtual servers Test environment management Containers Service virtualization INCIDENTS / BUGS Bug tracking Defects Causal Analysis Traceability between features, issues and code TEST MANAGEMENT Basic test management Test planning with development Testing before development Impact analysis Agile management Testing and development as one unified team FUNCTIONAL Functionality inventory (backlog) Traceability between TCs and features TCs, checklists, ET sessions Test design techniques Tests are prioritized Long-term code coverage strategy AUTOMATED Unit API automated tests UI automation Unit with minimal coverage Unit, API and UI automated tests running continuously PERFORMANCE Client-side performance tests System performance (before go-live) Reactive monitoring in production Unit performance tests (during development) Proactive monitoring in production Performance tests running continuously SECURITY Basic of access control policies OWASP top 10 Pen Automated security checks USABILITY Usability User Accessibility MANDATORY RECOMMENDED OPTIONAL

4 BASIC TESTING EFFICIENT TESTING CONTINUOUS TESTING SOURCE CODE Pieces of source code get lost. Not clear what version each client has, which makes it complicated to do fixes in the corresponding code. The code has a big technical debt, maintainability problems, poor internal quality, lack of documentation, dead or duplicated code, doesn't follow best practices in design or architecture, complex code (spaghetti), etc. Finding bugs and solving issues takes too long. Integration is complex and costly. ENVIRONMENT / INFRASTRUCTURE INCIDENTS / BUGS TEST MANAGEMENT Not clear what is in each environment, everyone works in shared environments. Not sure if we are with the latest version. Bad communication between development and. No knowledge of the state of each incident. No knowledge of the version in which an incident was fixed. No test cycles defined. Testing is hard, not business focused, starts late, and takes too long to update a test case. No knowledge of which incidents each test case corresponds to. Not clear what needs to be tested or when. Data is overwritten between developers, testers or automated tests. There are devices that have problems. Cannot test on all devices. No knowledge of how to avoid incidents. No knowledge of where the incidents come from. Testing starts after development, focused on detecting and reporting, not prevention. When something changes, no knowledge of which test cases need to be executed. Difficult to set up a new environment for a demo, test or whatever is necessary. No knowledge of which feature is affected by a certain bug and to what line of code it relates to. No traceability within code versions. Gap between development and team, not sharing goals. PAINS TO SOLVE FUNCTIONAL No record of what has to be tested or with which level of priority. No evidence of test executions. No information on the quality status of each version. Uncertain about how well the tests are designed. Not clear what to test first. No knowledge of what coverage we should have. Not enough time to meet the expected coverage. AUTOMATED Incidents already solved reappear. Getting feedback after introducing a new change takes too long. Automated tests take a long time to run. Automated tests are expensive in terms of maintenance. Testers are bored and demotivated, always executing the same tests. Regression tests are executed manually and take a long time. Testers make mistakes when doing checkups. Fear and uncertainty when releasing a new feature to production. PERFORMANCE Uncertainty when going live, lack of knowledge about how the system will perform. No control over production systems or other environments. No clear methodology to carry out tests that simulate the expected load. Performance problems are difficult to solve and are detected very late. Unable to anticipate problems that occur in production. No knowledge of how a new change affects performance. SECURITY Security breaches, uncontrolled risks or uncertainty concerning how unprotected the users are. Security standards are not met. No knowledge of how a new change affects security. Need to release frequent security patches. USABILITY Users find the system difficult to use. No evidence that the application is usable. Users are resistant to change due their lack of involvement in acceptance. No evidence that the application is user-friendly. No evidence that the application is accessible to all.

5 WHY CONTINUOUS DELIVERY? CASE STUDY By using continuous delivery practices, HP LaserJet Firmware team could: Reduce overall development costs by ~40% Increase programs under development by ~140% Reduce development costs per program by 78% Increase resources driving innovation by 5x Source: Thoughtworks - The Case for Continuous Delivery.